Building a DiscoGAN model

The base datasets in this problem are obtained from the edges2handbags (https://people.eecs.berkeley.edu/~tinghuiz/projects/pix2pix/datasets/edges2handbags.tar.gz) and edges2shoes (https://people.eecs.berkeley.edu/~tinghuiz/projects/pix2pix/datasets/edges2shoes.tar.gz) datasets. Each image that's present in these datasets contain two sub-images. One is the colored image of the object, while the other is the image of the edges of the corresponding color image.  

Follow the steps to build a DiscoGAN model:

1. First, resize and crop the images in this dataset to obtain the handbag and shoe images:

defextract_files(data_dir,type = 'bags'):
   '''
 :param data_dir: Input directory
 :param type: bags or shoes
   :return: saves the cropped files to the bags to shoes directory
   '''

   input_file_dir = os.path.join(os.getcwd(),data_dir, "train")
   result_dir = os.path.join(os.getcwd(),type)
if not os.path.exists(result_dir):
       os.makedirs(result_dir)

   file_names= os...